Application of multiattribute theory in a safety monitor for the planning of maintenance jobs

In the chemical industry periodic maintenance of the plant ususally requires an extensive and detailed plannning of the work that has to be completed in the rather short timespan of a production shutdown. Heavy workload and pressure of time often create conditions in which workers are exposed to considerable safety risks. A model has been developed that computes a danger measure for each maintenance job in the planning. This measure is in the form of a multiattribute value function which has an additive representation of properly scaled simple functions of single attributes. In the paper special attention is given to the problem of assessment of the scaling constants by experts. The method of weight allocation in a value tree is used for the elicitation of experts' judgements of safety threats during the execution of maintenance jobs. The computer model implemented is linked through a special interface with an existing planning system such that the model operates as a safety monitor for job schedules. The main results obtained with a prototype of the safety monitor are given in the paper.